Artificial Intelligence 2024: Breakthrough advances

Artificial intelligence (AI) is transforming our world at an astonishing rate. From chatbots that carry on smooth conversations to algorithms that outperform humans in solving complex problems, AI's achievements are becoming increasingly impressive. The "Artificial Intelligence Index Report 2024", published by the Institute for Human-Centred Artificial Intelligence at Stanford University, provides an overview of this dizzying progress and its implications.

The report highlights that artificial intelligence has surpassed humans in a number of metrics, such as image classification, visual reasoning and English language comprehension. However, it still lags behind in more complex tasks such as competition-level maths and common sense visual reasoning. As AI capabilities grow, existing benchmarks are rapidly becoming obsolete and new forms of assessment are needed.

Who is leading the AI race?

Industry continues to dominate cutting-edge AI research. In 2024, 51 known machine learning models are expected to be created by the industrial sector, while academia will only submit 15 models. The US continues to lead the way as the source of the best AI models: US institutions will create 61 known models, well ahead of the European Union with 21 models and China with 15 models. Technology giants such as OpenAI, Google and Anthropic are at the forefront of this rapid progress, playing a key role in the development and deployment of advanced AI technologies.

What are the costs and challenges of advanced AI?

As AI performance increases, so do the associated costs. It costs tens or even hundreds of millions of dollars to train next-generation models such as OpenAI's GPT-4 and Google's Gemini Ultra. These enormous costs include not only financial resources, but also the significant energy and water consumption required to keep these powerful systems running. The intensive use of natural resources raises serious environmental concerns.

In addition, the growing demand for high quality training data is becoming an increasingly tangible challenge. There is currently a limited amount of such data available, and the supply may run out in the coming years. This poses additional challenges for the further development of AI, as without sufficient data, the quality of training and performance of models may decrease significantly.

How does AI affect work and science?

Recent studies have shown that AI significantly improves labour productivity and the quality of work. These advanced systems can not only optimise processes, but also reduce the skills gap between low-skilled and high-skilled workers by providing access to tools and resources that were previously unavailable. In science, innovative AI applications such as Google's AlphaDev and GNoME are accelerating discovery and progress in a number of fields. These technologies are helping scientists analyse data with unprecedented speed and accuracy, opening up new opportunities for research and development.

How is the public responding to AI?

As technology enters the public consciousness, people around the world are becoming increasingly aware of its potential impact. A survey by Ipsos found that the proportion of people who believe AI will dramatically impact their lives in the next three to five years has increased from 60%to 66% in the last year. In addition, 52% express nervousness about AI products and services, up 13% from 2022.

Despite these challenges and concerns, enthusiasm for generative AI is growing. Investment in this field has grown nearly eightfold between 2023 and 2024, reaching 25.2 billion dollars. Major players in generative AI such as OpenAI, Anthropic, Hugging Face and Inflection have reported significant fundraising rounds. The Artificial Intelligence Index 2024 report paints a picture of rapid progress, huge potential, and complex challenges.

Growing Risks

Stanford's "Artificial Intelligence Index Report 2024" highlights the growing risks and challenges that accompany the rapid development of AI. A major challenge is the lack of reliable and standardised assessments needed to ensure the accountability and safety of AI models. Ethical issues also figure prominently in discussions surrounding the development and use of AI, including potential bias, privacy, transparency, and employment impacts.

The high cost and power consumption of advanced AI models call into question their sustainability and accessibility. To address these challenges, it is necessary to develop sound ethical principles, strengthen regulatory oversight, and encourage greater collaboration between industry, academia, and policymakers. These measures will help manage the growing risks and ensure the safe development of AI technologies.

While the report paints an impressive picture of the rapid development of AI, it also recognises the considerable uncertainty surrounding the future of this technology. The enormous potential of generative technologies has the potential to transform many aspects of society, from industry to healthcare. However, the exact trajectory and ultimate consequences of these changes are still difficult to predict. Addressing these questions about the future of AI will require continued research, dialogue and interdisciplinary collaboration. Its development and adoption will need to be actively managed, and the benefits of AI will need to be harnessed and challenges overcome.

ChatGPT and its use in the US

A recent study conducted by Pew Research Centre has shed light on the adoption and usage patterns of ChatGPT in the US. The study found that a significant portion of the adult population has experimented with this artificial intelligence tool, using it for a variety of personal and professional purposes.

15% of the US adult population, which is about 39 million people, have used ChatGPT. This level of prevalence is surprising given that the tool was launched relatively recently. Younger adults and those with higher levels of education are more likely to try ChatGPT. About 27% of adults under 30 and 29% of university graduates have used the tool, compared to 6% of those over 50 and 8% of those without a college degree.

Many Americans used ChatGPT for creative and writing tasks. About 46% of users used it to compose, edit, or generate ideas for various types of content, such as emails, essays, short stories, and poems. ChatGPT was also a popular resource for finding information and getting help with problems. About 44% of users used the tool to research specific topics or find answers to questions, and 31% used it for help with problems.

In addition to personal use, ChatGPT has found its way into professional and educational environments. Approximately 21% of users used the tool for work-related tasks, and 18% used it for educational or training purposes. A Pew Research Center study shows that ChatGPT has gained a significant user base in the United States, with young, educated adults at the forefront. Americans have found a wide range of uses for the tool, from content creation and information retrieval to professional and educational use. As ChatGPT and other artificial intelligence technologies such as Google Gemini and Meta AI develop, its impact on people's daily lives is likely to grow.

Conclusion

Artificial Intelligence in 2024 is showing astounding achievements and significant progress, confirming its ability to transform many aspects of our lives. We are on the cusp of new discoveries and changes, and how we meet these challenges and opportunities will determine the direction of AI for years to come.